Children with medulloblastoma and primitive neuroectodermal (PNET) brain tumors are divided into average- and high-risk groups so that therapy intensity can be tailored. Unfortunately, the methods for stratifying patients are imperfect, so that > 25% of children categorized as average risk fail therapy and die of disease. We have identified P27 loss as a biomarker that identifies most of these children and could be used to appropriately place them in the high-risk category, where they would receive intensified therapy increasing survival chances. Another clinical problem is that PNET patients are grouped with medulloblastoma patients in clinical trials and treated identically. We have recently learned that PNET tumors are molecularly distinct from medulloblastomas and that PNET patients respond poorly to therapy that has been tailored for medulloblastoma patients. The critical barrier to tailoring sPNET therapy to the disease has been the lack of relevant sPNET pre-clinical models. We have generated patient-derived sPNET stem cell cultures and orthotopic mouse models and have used these to identify FDA-approved drugs that are more efficacious in pre-clinical studies than currently used drugs. In this renewal application, our Specific Aims are: 1) To identify prognostic indicators of therapy failure for medulloblastoma and sPNET patients. 2) To prioritize targeted therapies for future clinical trials. The expected outcomes are: 1) that we will further refine molecular biomarkers suitable for patient stratification in a prospective analysis of > 300 patients; 2) that we will generate clinically relevant sPNET and high risk medulloblastoma pre-clinical models and 3) consistent with our track record, we will promote a regimen for human sPNET and high risk medulloblastoma trials based on studies that exceed all current standards for advancement of pre-clinical results into clinical trials. The significance is that we will ultimately abandon an archaic patient stratification scheme in favor o a biologically-based strategy that appropriately directs patients into tailored therapy and that sPNET patients will be treated with therapeutic regimens that are based on relevant pre-clinical and clinical data. Ultimately both will yield higher cure rates and limit toxicity in pediatric bran tumor patients.